When to Use Behavioral Economics Instead of Customer Research

Last updated October 9, 2026

When Microsoft tested product ideas its teams had already chosen to build, only about 1 in 3 improved the metric each idea was meant to move, compared with a control group that did not get the change. Ron Kohavi and colleagues reported that the rest made no measurable difference or made the metric worse. Ideas that seem right in interviews and planning meetings often fail once they meet real behavior, and that is where the choice between customer research and behavioral economics starts.

Use behavioral economics instead of customer research when you already understand what people need and the problem is what they do at the point of decision: they drop off partway, delay, stay with the default or stop following through. Customer research is the better tool for learning what people need and the words and circumstances they bring to a decision. At The Decision Lab, most of our projects use both in that order. Customer research tells us where and why behavior stalls, and a behavioral experiment against a control group tells us which change actually moves it.

That sequence is behind results like a fall by half in the dropout rate of a debt repayment program we redesigned for American Financial Solutions, and it runs through our more than 60 published case studies.

Behavioral economics is most useful when the problem is the gap between what people intend and what they do. Customer research is most useful when a team still needs to understand the customer, the problem, the context or the language.

Start with customer research when the open question is what people need. Start with behavioral economics when the open question is why people who want something do not do it.

To decide, work through these in order:

1. If you do not yet know what customers need, start with customer research.
2. If you know what they need and they are acting on it, keep researching the problem before designing any behavioral change.
3. If they fail to act despite stated interest or awareness, use behavioral economics to find and test the barriers in the decision.
4. If you already know the likely barrier, test a targeted change against a control group in a behavioral experiment. If you do not, use customer research to find it first, then test.

Customer research covers interviews, surveys, usability tests and field observation. It is how a team learns what people are trying to get done and the language and circumstances they bring to it.

Its limit is that it records what people notice and can report. Many of the forces that shape a decision sit in the design of the choice itself, such as which option is preselected or how many steps a form has, and people rarely mention them because they rarely notice them.

Behavioral economics is the use of behavioral science and experimental methods to improve real decisions, and much of it studies how those design features change what people do. Practitioners call the design of a choice its choice architecture, and the core skill is changing it and measuring the result. In our projects, customer research helps generate hypotheses about the barrier, and behavioral experiments test which intervention changes the outcome.

Each signal below describes a gap between what people want and what they do. When we see one in a client's data, we bring in behavioral economics alongside the customer research already under way.

1. Drop-off partway through a process

Use behavioral economics when people begin a process they want to finish and leave before the end, especially when effort or timing may be responsible.

Typical cases are sign-up flows and repayment plans. Customer research can confirm they meant to finish. It is less reliable on why they stopped, because the cause often sits earlier in the journey than the moment they quit.

American Financial Solutions, a nonprofit debt management program, asked us to reduce early dropout. Our modeling traced the risk back to elements of the first onboarding call, months before clients actually left. We redesigned the touchpoints, including counselor call scripts that draw out each client's own motivations at enrollment, and the dropout rate fell by half with the new touchpoints in place.

2. Low adoption despite awareness and interest

Use behavioral economics when people know about a product and say it would help, and usage still stays low.

A Fortune 500 HR technology company we worked with had deployed AI tools to a 20,000-person workforce with training and internal communications behind them, and usage stayed low.

We combined a review of the adoption literature with workforce interviews and surveys to find five behavioral barriers, including low trust in AI accuracy and poor fit with existing workflows. We then piloted eight programs for one month with more than 100 employees against a control group. Confidence in exploring AI tools, measured before and after, rose 41% among participants and fell 26% among control-group colleagues.

3. Poor adherence and follow-through

Use behavioral economics when people agree with a plan and then stop following it over time.

Typical cases are skipped doses and lapsed savings habits. The World Health Organization's 2003 report on long-term therapies estimated that adherence among patients with chronic diseases in developed countries averages about half, with lower rates in developing countries. The same report noted that patients are often blamed even though providers and health systems strongly influence whether they stick with treatment.

When knowledge and intention are already in place, asking people again what they want adds little. The useful work is in the timing of prompts and the effort each repeated action takes.

4. Status-quo bias: people stay with whatever is already in place

Use behavioral economics when outcomes depend on whether people change a setting or plan they already have.

When doing nothing is an option, many people take it, even when they would choose differently starting from scratch. William Samuelson and Richard Zeckhauser named this status-quo bias in 1988.

About 4 in 10 people chose to be donors when they had to opt in, compared with about 8 in 10 when they had to opt out, in Eric Johnson and Daniel Goldstein's organ donation default experiment with 161 participants. People were randomly assigned to the two versions, so their attitudes toward donation were the same on average, and a survey of those attitudes could not have told the two groups apart. When renewals or plan switches stall, the default is usually the first thing we test.

5. Complexity and effort

Use behavioral economics when people want the outcome and the steps to get it are long or confusing.

In a randomized field experiment on student aid applications, Eric Bettinger, Bridget Terry Long, Philip Oreopoulos and Lisa Sanbonmatsu had tax preparers help low-income families complete the US federal student aid application, using information already on their tax returns.

Among high school seniors whose parents got that help, the share who completed two years of college over the next three years was 36%, compared with 28% in the control group. The experiment separated information from application effort: personalized aid information on its own did not improve outcomes, while help completing the application, together with that information, did.

6. Trust barriers

Use behavioral economics together with customer research when people hold back because they doubt the product or the people behind it.

Trust is where the two methods depend most on each other, because customer research finds the specific reason and behavioral experiments test what restores reliance.

When we studied low adoption of clean cookstoves in Uganda for the World Bank, our field research identified 26 specific barriers. They included low trust and the weight of the stoves: the cheapest model was too heavy for most women to carry on their own.

On the testing side, Berkeley Dietvorst, Joseph Simmons and Cade Massey found that people stop relying on a forecasting algorithm after seeing it make a mistake, even after seeing it outperform a human forecaster. The same researchers later found that people use imperfect algorithms more readily when they can adjust the forecasts slightly.

We run customer research to find where and why behavior stalls, then run behavioral experiments to find the change that fixes it. In practice that means seven steps.

1. Name one behavior to change. Pick an observable action, such as completing enrollment within a week, rather than an attitude such as satisfaction.
2. Find where it breaks in your own data. Funnel analytics and usage logs show where people stop and how many of them stop there.
3. Run customer research at those points. Interviews, surveys, diary studies and observation explain what people were trying to do and what got in the way, in their own words.
4. Turn the findings into barrier hypotheses. Each hypothesis names a mechanism, such as effort or a default, and the group of people it affects.
5. Design more candidate changes than you will test. Each one should target one barrier for one group. For the Fortune 500 AI adoption project, we designed 20 programs and selected eight for the pilot.
6. Test against a control group. A randomized experiment gives the strongest evidence that a change caused a result. When randomization is not possible, or the change is cheap and easy to reverse, a staged rollout or a before-and-after comparison against a clear baseline can still guide the decision, as long as the team treats it as weaker evidence.
7. Scale what works and keep measuring. Results from a pilot can change when a program reaches everyone, so measurement continues after launch.

For Money Guided, an employee financial wellbeing product, we reviewed more than 50 studies and surveyed 533 UK employees and 264 HR managers before testing anything. We then ran a 400-participant controlled experiment comparing four framings of the product's value, wellness-led, finance-led, hybrid and competitor-style, against a control. The winning frame now shapes how the product presents itself to employees and to the HR managers who buy it. The published case study describes this as validation before launch, with behavior inside the product as the next test.

For a global consumer electronics company designing a mindfulness service inside a health app used by 60 million people, we built a design process that tested every target behavior through experimentation before production code was written. Design time fell 75%, even though the process added research steps the team had not run before.

Behavioral economics adds the most when there is an existing behavior to change and a way to measure it. Customer research on its own is usually the right call in these situations:

• You are defining a new product or category, and there is no behavior yet to observe.
• The question is about perception, such as how a brand is seen or which proposition people find easier to understand.
• The decision is a one-off strategic choice, such as which market to enter, with no repeated behavior to test.
• The audience is too small to compare groups, and qualitative depth is worth more than an underpowered experiment.

A common assumption is that high-stakes decisions, such as medical or financial ones, are too deliberate for behavioral factors to matter. The evidence points the other way: student aid applications moved with a change to effort alone, and the World Health Organization found that providers and health systems strongly influence whether patients stick with long-term treatment.

• Behavioral science vs. UX research vs. market research vs. management consulting compares the four disciplines and matches each to the business problems it fits best.
• How to choose a behavioral science consulting firm sets out ten questions to ask any firm before it runs experiments with your customers or employees.

What is the difference between behavioral economics and customer research?

Customer research asks people about their needs and experiences and records what they say and recall. Behavioral economics studies what people actually do and how the design of a choice, such as a default or the number of steps in a form, changes it. Customer research produces hypotheses about the barrier, and behavioral experiments test which change works.

Can behavioral economics replace customer research?

In most projects it should not. The same drop-off can come from effort, distrust, timing or a confusing default, and changes designed without customer research often target the wrong one. Customer research narrows the likely causes so experiments test the right ones. The exception is a problem where the barrier is already obvious from data, such as an opt-in step most people skip.

Do you need a randomized controlled trial to use behavioral economics?

A randomized controlled trial is not always necessary. It gives the strongest evidence that a change caused a result, and we use one whenever sample size and ethics allow. Otherwise, a staged rollout or a before-and-after comparison against a clear baseline can still guide a decision, provided the team treats it as weaker evidence and keeps measuring.

How long does a behavioral experiment take?

It depends on how often the behavior happens and how long the outcome takes to show. A sign-up flow with steady traffic can produce a result in weeks, while outcomes like college completion take years to observe. Our pilot of eight AI adoption programs at a Fortune 500 company ran for one month against a control group.